index-pack.c

   1 #include "cache.h"
   2 #include "delta.h"
   3 #include "pack.h"
   4 #include "csum-file.h"
   5 #include "blob.h"
   6 #include "commit.h"
   7 #include "tag.h"
   8 #include "tree.h"
   9 #include "progress.h"
  10 #include "fsck.h"
  11 #include "exec_cmd.h"
  12
  13 static const char index_pack_usage[] =
  14 "git index-pack [-v] [-o <index-file>] [{ ---keep | --keep=<msg> }] [--strict] { <pack-file> | --stdin [--fix-thin] [<pack-file>] }";
  15
  16 struct object_entry
  17 {
  18         struct pack_idx_entry idx;
  19         unsigned long size;
  20         unsigned int hdr_size;
  21         enum object_type type;
  22         enum object_type real_type;
  23 };
  24
  25 union delta_base {
  26         unsigned char sha1[20];
  27         off_t offset;
  28 };
  29
  30 struct base_data {
  31         struct base_data *base;
  32         struct base_data *child;
  33         struct object_entry *obj;
  34         void *data;
  35         unsigned long size;
  36 };
  37
  38 /*
  39  * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
  40  * to memcmp() only the first 20 bytes.
  41  */
  42 #define UNION_BASE_SZ   20
  43
  44 #define FLAG_LINK (1u<<20)
  45 #define FLAG_CHECKED (1u<<21)
  46
  47 struct delta_entry
  48 {
  49         union delta_base base;
  50         int obj_no;
  51 };
  52
  53 static struct object_entry *objects;
  54 static struct delta_entry *deltas;
  55 static struct base_data *base_cache;
  56 static size_t base_cache_used;
  57 static int nr_objects;
  58 static int nr_deltas;
  59 static int nr_resolved_deltas;
  60
  61 static int from_stdin;
  62 static int strict;
  63 static int verbose;
  64
  65 static struct progress *progress;
  66
  67 /* We always read in 4kB chunks. */
  68 static unsigned char input_buffer[4096];
  69 static unsigned int input_offset, input_len;
  70 static off_t consumed_bytes;
  71 static git_SHA_CTX input_ctx;
  72 static uint32_t input_crc32;
  73 static int input_fd, output_fd, pack_fd;
  74
  75 static int mark_link(struct object *obj, int type, void *data)
  76 {
  77         if (!obj)
  78                 return -1;
  79
  80         if (type != OBJ_ANY && obj->type != type)
  81                 die("object type mismatch at %s", sha1_to_hex(obj->sha1));
  82
  83         obj->flags |= FLAG_LINK;
  84         return 0;
  85 }
  86
  87 /* The content of each linked object must have been checked
  88    or it must be already present in the object database */
  89 static void check_object(struct object *obj)
  90 {
  91         if (!obj)
  92                 return;
  93
  94         if (!(obj->flags & FLAG_LINK))
  95                 return;
  96
  97         if (!(obj->flags & FLAG_CHECKED)) {
  98                 unsigned long size;
  99                 int type = sha1_object_info(obj->sha1, &size);
 100                 if (type != obj->type || type <= 0)
 101                         die("object of unexpected type");
 102                 obj->flags |= FLAG_CHECKED;
 103                 return;
 104         }
 105 }
 106
 107 static void check_objects(void)
 108 {
 109         unsigned i, max;
 110
 111         max = get_max_object_index();
 112         for (i = 0; i < max; i++)
 113                 check_object(get_indexed_object(i));
 114 }
 115
 116
 117 /* Discard current buffer used content. */
 118 static void flush(void)
 119 {
 120         if (input_offset) {
 121                 if (output_fd >= 0)
 122                         write_or_die(output_fd, input_buffer, input_offset);
 123                 git_SHA1_Update(&input_ctx, input_buffer, input_offset);
 124                 memmove(input_buffer, input_buffer + input_offset, input_len);
 125                 input_offset = 0;
 126         }
 127 }
 128
 129 /*
 130  * Make sure at least "min" bytes are available in the buffer, and
 131  * return the pointer to the buffer.
 132  */
 133 static void *fill(int min)
 134 {
 135         if (min <= input_len)
 136                 return input_buffer + input_offset;
 137         if (min > sizeof(input_buffer))
 138                 die("cannot fill %d bytes", min);
 139         flush();
 140         do {
 141                 ssize_t ret = xread(input_fd, input_buffer + input_len,
 142                                 sizeof(input_buffer) - input_len);
 143                 if (ret <= 0) {
 144                         if (!ret)
 145                                 die("early EOF");
 146                         die("read error on input: %s", strerror(errno));
 147                 }
 148                 input_len += ret;
 149                 if (from_stdin)
 150                         display_throughput(progress, consumed_bytes + input_len);
 151         } while (input_len < min);
 152         return input_buffer;
 153 }
 154
 155 static void use(int bytes)
 156 {
 157         if (bytes > input_len)
 158                 die("used more bytes than were available");
 159         input_crc32 = crc32(input_crc32, input_buffer + input_offset, bytes);
 160         input_len -= bytes;
 161         input_offset += bytes;
 162
 163         /* make sure off_t is sufficiently large not to wrap */
 164         if (consumed_bytes > consumed_bytes + bytes)
 165                 die("pack too large for current definition of off_t");
 166         consumed_bytes += bytes;
 167 }
 168
 169 static char *open_pack_file(char *pack_name)
 170 {
 171         if (from_stdin) {
 172                 input_fd = 0;
 173                 if (!pack_name) {
 174                         static char tmpfile[PATH_MAX];
 175                         snprintf(tmpfile, sizeof(tmpfile),
 176                                  "%s/pack/tmp_pack_XXXXXX", get_object_directory());
 177                         output_fd = xmkstemp(tmpfile);
 178                         pack_name = xstrdup(tmpfile);
 179                 } else
 180                         output_fd = open(pack_name, O_CREAT|O_EXCL|O_RDWR, 0600);
 181                 if (output_fd < 0)
 182                         die("unable to create %s: %s", pack_name, strerror(errno));
 183                 pack_fd = output_fd;
 184         } else {
 185                 input_fd = open(pack_name, O_RDONLY);
 186                 if (input_fd < 0)
 187                         die("cannot open packfile '%s': %s",
 188                             pack_name, strerror(errno));
 189                 output_fd = -1;
 190                 pack_fd = input_fd;
 191         }
 192         git_SHA1_Init(&input_ctx);
 193         return pack_name;
 194 }
 195
 196 static void parse_pack_header(void)
 197 {
 198         struct pack_header *hdr = fill(sizeof(struct pack_header));
 199
 200         /* Header consistency check */
 201         if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
 202                 die("pack signature mismatch");
 203         if (!pack_version_ok(hdr->hdr_version))
 204                 die("pack version %"PRIu32" unsupported",
 205                         ntohl(hdr->hdr_version));
 206
 207         nr_objects = ntohl(hdr->hdr_entries);
 208         use(sizeof(struct pack_header));
 209 }
 210
 211 static void bad_object(unsigned long offset, const char *format,
 212                        ...) NORETURN __attribute__((format (printf, 2, 3)));
 213
 214 static void bad_object(unsigned long offset, const char *format, ...)
 215 {
 216         va_list params;
 217         char buf[1024];
 218
 219         va_start(params, format);
 220         vsnprintf(buf, sizeof(buf), format, params);
 221         va_end(params);
 222         die("pack has bad object at offset %lu: %s", offset, buf);
 223 }
 224
 225 static void free_base_data(struct base_data *c)
 226 {
 227         if (c->data) {
 228                 free(c->data);
 229                 c->data = NULL;
 230                 base_cache_used -= c->size;
 231         }
 232 }
 233
 234 static void prune_base_data(struct base_data *retain)
 235 {
 236         struct base_data *b = base_cache;
 237         for (b = base_cache;
 238              base_cache_used > delta_base_cache_limit && b;
 239              b = b->child) {
 240                 if (b->data && b != retain)
 241                         free_base_data(b);
 242         }
 243 }
 244
 245 static void link_base_data(struct base_data *base, struct base_data *c)
 246 {
 247         if (base)
 248                 base->child = c;
 249         else
 250                 base_cache = c;
 251
 252         c->base = base;
 253         c->child = NULL;
 254         if (c->data)
 255                 base_cache_used += c->size;
 256         prune_base_data(c);
 257 }
 258
 259 static void unlink_base_data(struct base_data *c)
 260 {
 261         struct base_data *base = c->base;
 262         if (base)
 263                 base->child = NULL;
 264         else
 265                 base_cache = NULL;
 266         free_base_data(c);
 267 }
 268
 269 static void *unpack_entry_data(unsigned long offset, unsigned long size)
 270 {
 271         z_stream stream;
 272         void *buf = xmalloc(size);
 273
 274         memset(&stream, 0, sizeof(stream));
 275         stream.next_out = buf;
 276         stream.avail_out = size;
 277         stream.next_in = fill(1);
 278         stream.avail_in = input_len;
 279         git_inflate_init(&stream);
 280
 281         for (;;) {
 282                 int ret = git_inflate(&stream, 0);
 283                 use(input_len - stream.avail_in);
 284                 if (stream.total_out == size && ret == Z_STREAM_END)
 285                         break;
 286                 if (ret != Z_OK)
 287                         bad_object(offset, "inflate returned %d", ret);
 288                 stream.next_in = fill(1);
 289                 stream.avail_in = input_len;
 290         }
 291         git_inflate_end(&stream);
 292         return buf;
 293 }
 294
 295 static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 296 {
 297         unsigned char *p, c;
 298         unsigned long size;
 299         off_t base_offset;
 300         unsigned shift;
 301         void *data;
 302
 303         obj->idx.offset = consumed_bytes;
 304         input_crc32 = crc32(0, Z_NULL, 0);
 305
 306         p = fill(1);
 307         c = *p;
 308         use(1);
 309         obj->type = (c >> 4) & 7;
 310         size = (c & 15);
 311         shift = 4;
 312         while (c & 0x80) {
 313                 p = fill(1);
 314                 c = *p;
 315                 use(1);
 316                 size += (c & 0x7fUL) << shift;
 317                 shift += 7;
 318         }
 319         obj->size = size;
 320
 321         switch (obj->type) {
 322         case OBJ_REF_DELTA:
 323                 hashcpy(delta_base->sha1, fill(20));
 324                 use(20);
 325                 break;
 326         case OBJ_OFS_DELTA:
 327                 memset(delta_base, 0, sizeof(*delta_base));
 328                 p = fill(1);
 329                 c = *p;
 330                 use(1);
 331                 base_offset = c & 127;
 332                 while (c & 128) {
 333                         base_offset += 1;
 334                         if (!base_offset || MSB(base_offset, 7))
 335                                 bad_object(obj->idx.offset, "offset value overflow for delta base object");
 336                         p = fill(1);
 337                         c = *p;
 338                         use(1);
 339                         base_offset = (base_offset << 7) + (c & 127);
 340                 }
 341                 delta_base->offset = obj->idx.offset - base_offset;
 342                 if (delta_base->offset <= 0 || delta_base->offset >= obj->idx.offset)
 343                         bad_object(obj->idx.offset, "delta base offset is out of bound");
 344                 break;
 345         case OBJ_COMMIT:
 346         case OBJ_TREE:
 347         case OBJ_BLOB:
 348         case OBJ_TAG:
 349                 break;
 350         default:
 351                 bad_object(obj->idx.offset, "unknown object type %d", obj->type);
 352         }
 353         obj->hdr_size = consumed_bytes - obj->idx.offset;
 354
 355         data = unpack_entry_data(obj->idx.offset, obj->size);
 356         obj->idx.crc32 = input_crc32;
 357         return data;
 358 }
 359
 360 static void *get_data_from_pack(struct object_entry *obj)
 361 {
 362         off_t from = obj[0].idx.offset + obj[0].hdr_size;
 363         unsigned long len = obj[1].idx.offset - from;
 364         unsigned long rdy = 0;
 365         unsigned char *src, *data;
 366         z_stream stream;
 367         int st;
 368
 369         src = xmalloc(len);
 370         data = src;
 371         do {
 372                 ssize_t n = pread(pack_fd, data + rdy, len - rdy, from + rdy);
 373                 if (n < 0)
 374                         die("cannot pread pack file: %s", strerror(errno));
 375                 if (!n)
 376                         die("premature end of pack file, %lu bytes missing",
 377                             len - rdy);
 378                 rdy += n;
 379         } while (rdy < len);
 380         data = xmalloc(obj->size);
 381         memset(&stream, 0, sizeof(stream));
 382         stream.next_out = data;
 383         stream.avail_out = obj->size;
 384         stream.next_in = src;
 385         stream.avail_in = len;
 386         git_inflate_init(&stream);
 387         while ((st = git_inflate(&stream, Z_FINISH)) == Z_OK);
 388         git_inflate_end(&stream);
 389         if (st != Z_STREAM_END || stream.total_out != obj->size)
 390                 die("serious inflate inconsistency");
 391         free(src);
 392         return data;
 393 }
 394
 395 static int find_delta(const union delta_base *base)
 396 {
 397         int first = 0, last = nr_deltas;
 398
 399         while (first < last) {
 400                 int next = (first + last) / 2;
 401                 struct delta_entry *delta = &deltas[next];
 402                 int cmp;
 403
 404                 cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
 405                 if (!cmp)
 406                         return next;
 407                 if (cmp < 0) {
 408                         last = next;
 409                         continue;
 410                 }
 411                 first = next+1;
 412         }
 413         return -first-1;
 414 }
 415
 416 static void find_delta_children(const union delta_base *base,
 417                                 int *first_index, int *last_index)
 418 {
 419         int first = find_delta(base);
 420         int last = first;
 421         int end = nr_deltas - 1;
 422
 423         if (first < 0) {
 424                 *first_index = 0;
 425                 *last_index = -1;
 426                 return;
 427         }
 428         while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
 429                 --first;
 430         while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
 431                 ++last;
 432         *first_index = first;
 433         *last_index = last;
 434 }
 435
 436 static void sha1_object(const void *data, unsigned long size,
 437                         enum object_type type, unsigned char *sha1)
 438 {
 439         hash_sha1_file(data, size, typename(type), sha1);
 440         if (has_sha1_file(sha1)) {
 441                 void *has_data;
 442                 enum object_type has_type;
 443                 unsigned long has_size;
 444                 has_data = read_sha1_file(sha1, &has_type, &has_size);
 445                 if (!has_data)
 446                         die("cannot read existing object %s", sha1_to_hex(sha1));
 447                 if (size != has_size || type != has_type ||
 448                     memcmp(data, has_data, size) != 0)
 449                         die("SHA1 COLLISION FOUND WITH %s !", sha1_to_hex(sha1));
 450                 free(has_data);
 451         }
 452         if (strict) {
 453                 if (type == OBJ_BLOB) {
 454                         struct blob *blob = lookup_blob(sha1);
 455                         if (blob)
 456                                 blob->object.flags |= FLAG_CHECKED;
 457                         else
 458                                 die("invalid blob object %s", sha1_to_hex(sha1));
 459                 } else {
 460                         struct object *obj;
 461                         int eaten;
 462                         void *buf = (void *) data;
 463
 464                         /*
 465                          * we do not need to free the memory here, as the
 466                          * buf is deleted by the caller.
 467                          */
 468                         obj = parse_object_buffer(sha1, type, size, buf, &eaten);
 469                         if (!obj)
 470                                 die("invalid %s", typename(type));
 471                         if (fsck_object(obj, 1, fsck_error_function))
 472                                 die("Error in object");
 473                         if (fsck_walk(obj, mark_link, 0))
 474                                 die("Not all child objects of %s are reachable", sha1_to_hex(obj->sha1));
 475
 476                         if (obj->type == OBJ_TREE) {
 477                                 struct tree *item = (struct tree *) obj;
 478                                 item->buffer = NULL;
 479                         }
 480                         if (obj->type == OBJ_COMMIT) {
 481                                 struct commit *commit = (struct commit *) obj;
 482                                 commit->buffer = NULL;
 483                         }
 484                         obj->flags |= FLAG_CHECKED;
 485                 }
 486         }
 487 }
 488
 489 static void *get_base_data(struct base_data *c)
 490 {
 491         if (!c->data) {
 492                 struct object_entry *obj = c->obj;
 493
 494                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 495                         void *base = get_base_data(c->base);
 496                         void *raw = get_data_from_pack(obj);
 497                         c->data = patch_delta(
 498                                 base, c->base->size,
 499                                 raw, obj->size,
 500                                 &c->size);
 501                         free(raw);
 502                         if (!c->data)
 503                                 bad_object(obj->idx.offset, "failed to apply delta");
 504                 } else {
 505                         c->data = get_data_from_pack(obj);
 506                         c->size = obj->size;
 507                 }
 508
 509                 base_cache_used += c->size;
 510                 prune_base_data(c);
 511         }
 512         return c->data;
 513 }
 514
 515 static void resolve_delta(struct object_entry *delta_obj,
 516                           struct base_data *base, struct base_data *result)
 517 {
 518         void *base_data, *delta_data;
 519
 520         delta_obj->real_type = base->obj->real_type;
 521         delta_data = get_data_from_pack(delta_obj);
 522         base_data = get_base_data(base);
 523         result->obj = delta_obj;
 524         result->data = patch_delta(base_data, base->size,
 525                                    delta_data, delta_obj->size, &result->size);
 526         free(delta_data);
 527         if (!result->data)
 528                 bad_object(delta_obj->idx.offset, "failed to apply delta");
 529         sha1_object(result->data, result->size, delta_obj->real_type,
 530                     delta_obj->idx.sha1);
 531         nr_resolved_deltas++;
 532 }
 533
 534 static void find_unresolved_deltas(struct base_data *base,
 535                                    struct base_data *prev_base)
 536 {
 537         int i, ref_first, ref_last, ofs_first, ofs_last;
 538
 539         /*
 540          * This is a recursive function. Those brackets should help reducing
 541          * stack usage by limiting the scope of the delta_base union.
 542          */
 543         {
 544                 union delta_base base_spec;
 545
 546                 hashcpy(base_spec.sha1, base->obj->idx.sha1);
 547                 find_delta_children(&base_spec, &ref_first, &ref_last);
 548
 549                 memset(&base_spec, 0, sizeof(base_spec));
 550                 base_spec.offset = base->obj->idx.offset;
 551                 find_delta_children(&base_spec, &ofs_first, &ofs_last);
 552         }
 553
 554         if (ref_last == -1 && ofs_last == -1) {
 555                 free(base->data);
 556                 return;
 557         }
 558
 559         link_base_data(prev_base, base);
 560
 561         for (i = ref_first; i <= ref_last; i++) {
 562                 struct object_entry *child = objects + deltas[i].obj_no;
 563                 if (child->real_type == OBJ_REF_DELTA) {
 564                         struct base_data result;
 565                         resolve_delta(child, base, &result);
 566                         if (i == ref_last && ofs_last == -1)
 567                                 free_base_data(base);
 568                         find_unresolved_deltas(&result, base);
 569                 }
 570         }
 571
 572         for (i = ofs_first; i <= ofs_last; i++) {
 573                 struct object_entry *child = objects + deltas[i].obj_no;
 574                 if (child->real_type == OBJ_OFS_DELTA) {
 575                         struct base_data result;
 576                         resolve_delta(child, base, &result);
 577                         if (i == ofs_last)
 578                                 free_base_data(base);
 579                         find_unresolved_deltas(&result, base);
 580                 }
 581         }
 582
 583         unlink_base_data(base);
 584 }
 585
 586 static int compare_delta_entry(const void *a, const void *b)
 587 {
 588         const struct delta_entry *delta_a = a;
 589         const struct delta_entry *delta_b = b;
 590         return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 591 }
 592
 593 /* Parse all objects and return the pack content SHA1 hash */
 594 static void parse_pack_objects(unsigned char *sha1)
 595 {
 596         int i;
 597         struct delta_entry *delta = deltas;
 598         struct stat st;
 599
 600         /*
 601          * First pass:
 602          * - find locations of all objects;
 603          * - calculate SHA1 of all non-delta objects;
 604          * - remember base (SHA1 or offset) for all deltas.
 605          */
 606         if (verbose)
 607                 progress = start_progress(
 608                                 from_stdin ? "Receiving objects" : "Indexing objects",
 609                                 nr_objects);
 610         for (i = 0; i < nr_objects; i++) {
 611                 struct object_entry *obj = &objects[i];
 612                 void *data = unpack_raw_entry(obj, &delta->base);
 613                 obj->real_type = obj->type;
 614                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
 615                         nr_deltas++;
 616                         delta->obj_no = i;
 617                         delta++;
 618                 } else
 619                         sha1_object(data, obj->size, obj->type, obj->idx.sha1);
 620                 free(data);
 621                 display_progress(progress, i+1);
 622         }
 623         objects[i].idx.offset = consumed_bytes;
 624         stop_progress(&progress);
 625
 626         /* Check pack integrity */
 627         flush();
 628         git_SHA1_Final(sha1, &input_ctx);
 629         if (hashcmp(fill(20), sha1))
 630                 die("pack is corrupted (SHA1 mismatch)");
 631         use(20);
 632
 633         /* If input_fd is a file, we should have reached its end now. */
 634         if (fstat(input_fd, &st))
 635                 die("cannot fstat packfile: %s", strerror(errno));
 636         if (S_ISREG(st.st_mode) &&
 637                         lseek(input_fd, 0, SEEK_CUR) - input_len != st.st_size)
 638                 die("pack has junk at the end");
 639
 640         if (!nr_deltas)
 641                 return;
 642
 643         /* Sort deltas by base SHA1/offset for fast searching */
 644         qsort(deltas, nr_deltas, sizeof(struct delta_entry),
 645               compare_delta_entry);
 646
 647         /*
 648          * Second pass:
 649          * - for all non-delta objects, look if it is used as a base for
 650          *   deltas;
 651          * - if used as a base, uncompress the object and apply all deltas,
 652          *   recursively checking if the resulting object is used as a base
 653          *   for some more deltas.
 654          */
 655         if (verbose)
 656                 progress = start_progress("Resolving deltas", nr_deltas);
 657         for (i = 0; i < nr_objects; i++) {
 658                 struct object_entry *obj = &objects[i];
 659                 struct base_data base_obj;
 660
 661                 if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
 662                         continue;
 663                 base_obj.obj = obj;
 664                 base_obj.data = NULL;
 665                 find_unresolved_deltas(&base_obj, NULL);
 666                 display_progress(progress, nr_resolved_deltas);
 667         }
 668 }
 669
 670 static int write_compressed(struct sha1file *f, void *in, unsigned int size)
 671 {
 672         z_stream stream;
 673         unsigned long maxsize;
 674         void *out;
 675
 676         memset(&stream, 0, sizeof(stream));
 677         deflateInit(&stream, zlib_compression_level);
 678         maxsize = deflateBound(&stream, size);
 679         out = xmalloc(maxsize);
 680
 681         /* Compress it */
 682         stream.next_in = in;
 683         stream.avail_in = size;
 684         stream.next_out = out;
 685         stream.avail_out = maxsize;
 686         while (deflate(&stream, Z_FINISH) == Z_OK);
 687         deflateEnd(&stream);
 688
 689         size = stream.total_out;
 690         sha1write(f, out, size);
 691         free(out);
 692         return size;
 693 }
 694
 695 static struct object_entry *append_obj_to_pack(struct sha1file *f,
 696                                const unsigned char *sha1, void *buf,
 697                                unsigned long size, enum object_type type)
 698 {
 699         struct object_entry *obj = &objects[nr_objects++];
 700         unsigned char header[10];
 701         unsigned long s = size;
 702         int n = 0;
 703         unsigned char c = (type << 4) | (s & 15);
 704         s >>= 4;
 705         while (s) {
 706                 header[n++] = c | 0x80;
 707                 c = s & 0x7f;
 708                 s >>= 7;
 709         }
 710         header[n++] = c;
 711         crc32_begin(f);
 712         sha1write(f, header, n);
 713         obj[0].size = size;
 714         obj[0].hdr_size = n;
 715         obj[0].type = type;
 716         obj[0].real_type = type;
 717         obj[1].idx.offset = obj[0].idx.offset + n;
 718         obj[1].idx.offset += write_compressed(f, buf, size);
 719         obj[0].idx.crc32 = crc32_end(f);
 720         sha1flush(f);
 721         hashcpy(obj->idx.sha1, sha1);
 722         return obj;
 723 }
 724
 725 static int delta_pos_compare(const void *_a, const void *_b)
 726 {
 727         struct delta_entry *a = *(struct delta_entry **)_a;
 728         struct delta_entry *b = *(struct delta_entry **)_b;
 729         return a->obj_no - b->obj_no;
 730 }
 731
 732 static void fix_unresolved_deltas(struct sha1file *f, int nr_unresolved)
 733 {
 734         struct delta_entry **sorted_by_pos;
 735         int i, n = 0;
 736
 737         /*
 738          * Since many unresolved deltas may well be themselves base objects
 739          * for more unresolved deltas, we really want to include the
 740          * smallest number of base objects that would cover as much delta
 741          * as possible by picking the
 742          * trunc deltas first, allowing for other deltas to resolve without
 743          * additional base objects.  Since most base objects are to be found
 744          * before deltas depending on them, a good heuristic is to start
 745          * resolving deltas in the same order as their position in the pack.
 746          */
 747         sorted_by_pos = xmalloc(nr_unresolved * sizeof(*sorted_by_pos));
 748         for (i = 0; i < nr_deltas; i++) {
 749                 if (objects[deltas[i].obj_no].real_type != OBJ_REF_DELTA)
 750                         continue;
 751                 sorted_by_pos[n++] = &deltas[i];
 752         }
 753         qsort(sorted_by_pos, n, sizeof(*sorted_by_pos), delta_pos_compare);
 754
 755         for (i = 0; i < n; i++) {
 756                 struct delta_entry *d = sorted_by_pos[i];
 757                 enum object_type type;
 758                 struct base_data base_obj;
 759
 760                 if (objects[d->obj_no].real_type != OBJ_REF_DELTA)
 761                         continue;
 762                 base_obj.data = read_sha1_file(d->base.sha1, &type, &base_obj.size);
 763                 if (!base_obj.data)
 764                         continue;
 765
 766                 if (check_sha1_signature(d->base.sha1, base_obj.data,
 767                                 base_obj.size, typename(type)))
 768                         die("local object %s is corrupt", sha1_to_hex(d->base.sha1));
 769                 base_obj.obj = append_obj_to_pack(f, d->base.sha1,
 770                                         base_obj.data, base_obj.size, type);
 771                 find_unresolved_deltas(&base_obj, NULL);
 772                 display_progress(progress, nr_resolved_deltas);
 773         }
 774         free(sorted_by_pos);
 775 }
 776
 777 static void final(const char *final_pack_name, const char *curr_pack_name,
 778                   const char *final_index_name, const char *curr_index_name,
 779                   const char *keep_name, const char *keep_msg,
 780                   unsigned char *sha1)
 781 {
 782         const char *report = "pack";
 783         char name[PATH_MAX];
 784         int err;
 785
 786         if (!from_stdin) {
 787                 close(input_fd);
 788         } else {
 789                 fsync_or_die(output_fd, curr_pack_name);
 790                 err = close(output_fd);
 791                 if (err)
 792                         die("error while closing pack file: %s", strerror(errno));
 793         }
 794
 795         if (keep_msg) {
 796                 int keep_fd, keep_msg_len = strlen(keep_msg);
 797                 if (!keep_name) {
 798                         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 799                                  get_object_directory(), sha1_to_hex(sha1));
 800                         keep_name = name;
 801                 }
 802                 keep_fd = open(keep_name, O_RDWR|O_CREAT|O_EXCL, 0600);
 803                 if (keep_fd < 0) {
 804                         if (errno != EEXIST)
 805                                 die("cannot write keep file");
 806                 } else {
 807                         if (keep_msg_len > 0) {
 808                                 write_or_die(keep_fd, keep_msg, keep_msg_len);
 809                                 write_or_die(keep_fd, "\n", 1);
 810                         }
 811                         if (close(keep_fd) != 0)
 812                                 die("cannot write keep file");
 813                         report = "keep";
 814                 }
 815         }
 816
 817         if (final_pack_name != curr_pack_name) {
 818                 if (!final_pack_name) {
 819                         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 820                                  get_object_directory(), sha1_to_hex(sha1));
 821                         final_pack_name = name;
 822                 }
 823                 if (move_temp_to_file(curr_pack_name, final_pack_name))
 824                         die("cannot store pack file");
 825         }
 826         if (from_stdin)
 827                 chmod(final_pack_name, 0444);
 828
 829         if (final_index_name != curr_index_name) {
 830                 if (!final_index_name) {
 831                         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 832                                  get_object_directory(), sha1_to_hex(sha1));
 833                         final_index_name = name;
 834                 }
 835                 if (move_temp_to_file(curr_index_name, final_index_name))
 836                         die("cannot store index file");
 837         }
 838         chmod(final_index_name, 0444);
 839
 840         if (!from_stdin) {
 841                 printf("%s\n", sha1_to_hex(sha1));
 842         } else {
 843                 char buf[48];
 844                 int len = snprintf(buf, sizeof(buf), "%s\t%s\n",
 845                                    report, sha1_to_hex(sha1));
 846                 write_or_die(1, buf, len);
 847
 848                 /*
 849                  * Let's just mimic git-unpack-objects here and write
 850                  * the last part of the input buffer to stdout.
 851                  */
 852                 while (input_len) {
 853                         err = xwrite(1, input_buffer + input_offset, input_len);
 854                         if (err <= 0)
 855                                 break;
 856                         input_len -= err;
 857                         input_offset += err;
 858                 }
 859         }
 860 }
 861
 862 static int git_index_pack_config(const char *k, const char *v, void *cb)
 863 {
 864         if (!strcmp(k, "pack.indexversion")) {
 865                 pack_idx_default_version = git_config_int(k, v);
 866                 if (pack_idx_default_version > 2)
 867                         die("bad pack.indexversion=%"PRIu32,
 868                                 pack_idx_default_version);
 869                 return 0;
 870         }
 871         return git_default_config(k, v, cb);
 872 }
 873
 874 int main(int argc, char **argv)
 875 {
 876         int i, fix_thin_pack = 0;
 877         char *curr_pack, *pack_name = NULL;
 878         char *curr_index, *index_name = NULL;
 879         const char *keep_name = NULL, *keep_msg = NULL;
 880         char *index_name_buf = NULL, *keep_name_buf = NULL;
 881         struct pack_idx_entry **idx_objects;
 882         unsigned char pack_sha1[20];
 883
 884         git_extract_argv0_path(argv[0]);
 885
 886         /*
 887          * We wish to read the repository's config file if any, and
 888          * for that it is necessary to call setup_git_directory_gently().
 889          * However if the cwd was inside .git/objects/pack/ then we need
 890          * to go back there or all the pack name arguments will be wrong.
 891          * And in that case we cannot rely on any prefix returned by
 892          * setup_git_directory_gently() either.
 893          */
 894         {
 895                 char cwd[PATH_MAX+1];
 896                 int nongit;
 897
 898                 if (!getcwd(cwd, sizeof(cwd)-1))
 899                         die("Unable to get current working directory");
 900                 setup_git_directory_gently(&nongit);
 901                 git_config(git_index_pack_config, NULL);
 902                 if (chdir(cwd))
 903                         die("Cannot come back to cwd");
 904         }
 905
 906         for (i = 1; i < argc; i++) {
 907                 char *arg = argv[i];
 908
 909                 if (*arg == '-') {
 910                         if (!strcmp(arg, "--stdin")) {
 911                                 from_stdin = 1;
 912                         } else if (!strcmp(arg, "--fix-thin")) {
 913                                 fix_thin_pack = 1;
 914                         } else if (!strcmp(arg, "--strict")) {
 915                                 strict = 1;
 916                         } else if (!strcmp(arg, "--keep")) {
 917                                 keep_msg = "";
 918                         } else if (!prefixcmp(arg, "--keep=")) {
 919                                 keep_msg = arg + 7;
 920                         } else if (!prefixcmp(arg, "--pack_header=")) {
 921                                 struct pack_header *hdr;
 922                                 char *c;
 923
 924                                 hdr = (struct pack_header *)input_buffer;
 925                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 926                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 927                                 if (*c != ',')
 928                                         die("bad %s", arg);
 929                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 930                                 if (*c)
 931                                         die("bad %s", arg);
 932                                 input_len = sizeof(*hdr);
 933                         } else if (!strcmp(arg, "-v")) {
 934                                 verbose = 1;
 935                         } else if (!strcmp(arg, "-o")) {
 936                                 if (index_name || (i+1) >= argc)
 937                                         usage(index_pack_usage);
 938                                 index_name = argv[++i];
 939                         } else if (!prefixcmp(arg, "--index-version=")) {
 940                                 char *c;
 941                                 pack_idx_default_version = strtoul(arg + 16, &c, 10);
 942                                 if (pack_idx_default_version > 2)
 943                                         die("bad %s", arg);
 944                                 if (*c == ',')
 945                                         pack_idx_off32_limit = strtoul(c+1, &c, 0);
 946                                 if (*c || pack_idx_off32_limit & 0x80000000)
 947                                         die("bad %s", arg);
 948                         } else
 949                                 usage(index_pack_usage);
 950                         continue;
 951                 }
 952
 953                 if (pack_name)
 954                         usage(index_pack_usage);
 955                 pack_name = arg;
 956         }
 957
 958         if (!pack_name && !from_stdin)
 959                 usage(index_pack_usage);
 960         if (fix_thin_pack && !from_stdin)
 961                 die("--fix-thin cannot be used without --stdin");
 962         if (!index_name && pack_name) {
 963                 int len = strlen(pack_name);
 964                 if (!has_extension(pack_name, ".pack"))
 965                         die("packfile name '%s' does not end with '.pack'",
 966                             pack_name);
 967                 index_name_buf = xmalloc(len);
 968                 memcpy(index_name_buf, pack_name, len - 5);
 969                 strcpy(index_name_buf + len - 5, ".idx");
 970                 index_name = index_name_buf;
 971         }
 972         if (keep_msg && !keep_name && pack_name) {
 973                 int len = strlen(pack_name);
 974                 if (!has_extension(pack_name, ".pack"))
 975                         die("packfile name '%s' does not end with '.pack'",
 976                             pack_name);
 977                 keep_name_buf = xmalloc(len);
 978                 memcpy(keep_name_buf, pack_name, len - 5);
 979                 strcpy(keep_name_buf + len - 5, ".keep");
 980                 keep_name = keep_name_buf;
 981         }
 982
 983         curr_pack = open_pack_file(pack_name);
 984         parse_pack_header();
 985         objects = xmalloc((nr_objects + 1) * sizeof(struct object_entry));
 986         deltas = xmalloc(nr_objects * sizeof(struct delta_entry));
 987         parse_pack_objects(pack_sha1);
 988         if (nr_deltas == nr_resolved_deltas) {
 989                 stop_progress(&progress);
 990                 /* Flush remaining pack final 20-byte SHA1. */
 991                 flush();
 992         } else {
 993                 if (fix_thin_pack) {
 994                         struct sha1file *f;
 995                         unsigned char read_sha1[20], tail_sha1[20];
 996                         char msg[48];
 997                         int nr_unresolved = nr_deltas - nr_resolved_deltas;
 998                         int nr_objects_initial = nr_objects;
 999                         if (nr_unresolved <= 0)
1000                                 die("confusion beyond insanity");
1001                         objects = xrealloc(objects,
1002                                            (nr_objects + nr_unresolved + 1)
1003                                            * sizeof(*objects));
1004                         f = sha1fd(output_fd, curr_pack);
1005                         fix_unresolved_deltas(f, nr_unresolved);
1006                         sprintf(msg, "completed with %d local objects",
1007                                 nr_objects - nr_objects_initial);
1008                         stop_progress_msg(&progress, msg);
1009                         sha1close(f, tail_sha1, 0);
1010                         hashcpy(read_sha1, pack_sha1);
1011                         fixup_pack_header_footer(output_fd, pack_sha1,
1012                                                  curr_pack, nr_objects,
1013                                                  read_sha1, consumed_bytes-20);
1014                         if (hashcmp(read_sha1, tail_sha1) != 0)
1015                                 die("Unexpected tail checksum for %s "
1016                                     "(disk corruption?)", curr_pack);
1017                 }
1018                 if (nr_deltas != nr_resolved_deltas)
1019                         die("pack has %d unresolved deltas",
1020                             nr_deltas - nr_resolved_deltas);
1021         }
1022         free(deltas);
1023         if (strict)
1024                 check_objects();
1025
1026         idx_objects = xmalloc((nr_objects) * sizeof(struct pack_idx_entry *));
1027         for (i = 0; i < nr_objects; i++)
1028                 idx_objects[i] = &objects[i].idx;
1029         curr_index = write_idx_file(index_name, idx_objects, nr_objects, pack_sha1);
1030         free(idx_objects);
1031
1032         final(pack_name, curr_pack,
1033                 index_name, curr_index,
1034                 keep_name, keep_msg,
1035                 pack_sha1);
1036         free(objects);
1037         free(index_name_buf);
1038         free(keep_name_buf);
1039         if (pack_name == NULL)
1040                 free(curr_pack);
1041         if (index_name == NULL)
1042                 free(curr_index);
1043
1044         return 0;
1045 }